Brand tracker surveys developed to provide marketers with data and insight where none existed before. And it’s true that backward-looking, inevitably unrepresentative surveys are better than nothing when it comes to understanding your brand equity and making plans to manage it. However it’s vital that both marketers and researchers remember that traditional brand trackers are inherently a ‘make do’ solution. Our industry has been waiting its whole life for something better to come along. Now that something better has arrived.

Predicting brand equity

In an ideal world, nobody would ask a market researcher to tell them what their brand equity was a month or two ago. Neither would they ask to know the thoughts and feelings only of people who were prepared to spend time answering questions. Marketers want a full picture of what’s really happening to their brand, right now. And they want to know in time to do something about it. Yet that’s not what they currently get.

The next generation of tracking is ready 

For as long as ‘big data’ has been doing the rounds as a concept, there’s been the promise of using social media and search content to provide an instant, real-time snapshot of what consumers are saying, thinking and feeling. The potential is obvious: you have access to a full range of unfiltered opinions that respond to events as they happen and provide you with insights that are inherently actionable, in a timeframe that lets you do something about them. It helps too that mining the potential of search and social data is likely to be more cost-effective than enlisting an army of researchers to stop people in the street, call them on the phone or ping them with online surveys.

However, there have been plenty of understandable concerns about the real capability of search and social to perform the brand tracker role. On the one hand, the sheer noise of all the tweets, posts, updates and searches on the web makes isolating those that actually mean something to a particular brand look like a Herculean task. On the other, that megalith of social media and search content isn’t itself fully representative. Not everybody embraces social media with enthusiasm – what of the significant number of people who prefer to express their feelings and opinions the old-fashioned way? Do social media loudmouths who tweet 30 times a day inevitably outweigh those who restrict themselves to an hour or so of social media time in the evenings?

Representativeness has always been a central concern of researchers. However, if we are honest with ourselves, it’s a concern only because it affects our ability to predict the future accurately. We’re not in this business to give everyone an equal chance to have their say; we’re in it to reveal the real situation in as rapid and efficient a way as possible. The iron test isn’t whether search and social data captures the views of every single consumer out there; it’s whether we can accurately predict changes in brand equity on the basis of the consumers it does capture.

In an exhaustive series of studies covering 61 separate brands across a broad range of categories, TNS has proven that we can.

How social and search can predict the future

We developed an approach for aggregating and cleaning social media and search data to isolate only the content of genuine relevance to a brand. We also developed techniques for integrating this cleaned data with previous brand equity studies for that particular brand, to identify the big data signals that would correspond to shifts in brand equity. And then we tested how effective our search and social model would have been in predicting shifts in brand equity that had taken place in the past.

The results were nothing short of spectacular. Our social and search listening data delivered an R2 of 90% when it came to correlating with brand equity surveys. In other words, our approach explained 90% of the variations in brand equity that subsequently took place. And it did so up to 8 weeks in advance of when these changes appeared in survey results. By unleashing the power of big data, we were able to bring previously retrospective brand tracker insights into the real-time present, at a stroke.

Look deeper, and the results are more promising still. Mentions of brand attributes in social media correlated far more closely with actual brand equity than did the attributes listed in responses to survey questions. And, when marketing spend data is combined with social media data, modeled sales predictions are even more accurate. The fact that social media and search can predict brand equity isn’t a fluke – it’s a natural outcome of data that appears more robust and more predictive at every level.

  1. Mind and the machine: perfecting search and social listening 

How were we able to overcome the concerns that exist around search and social media listening, and deliver such an accurate prediction of brand equity? By integrating contextual understanding into the isolation, analysis and modeling of the data. We didn’t invite machines to do our thinking for us; instead we used intelligent analysis of the context for each brand to translate the stream of content into meaningful results.

When it comes to cleaning the data and isolating relevant content and searches, there are obvious steps involved such as targeting the right geographies, de-duping, removing the distortions that result from coupon activity and ensuring that content is genuinely about the category. When is a discussion about Apple related to technology and when is it concerned with baking or groceries, for example? However, there are also more contextual judgments to be made about when content is relevant – and what it really signifies: what are the relevant attributes for a category like technology, auto, or paper towels? Which types of comments relate to those attributes? And what constitutes a positive, negative or neutral comment in this context?

  1. Detecting the changing conversation 

The answers to these questions can change over time, of course. And in the real-time world that search and social media listening enables, they can change quite quickly. Because of this, the ability of our model to detect emerging trends and adjust to track them can be hugely important. The launch of a new product such as the Apple Watch produces rich listening data that can help marketers fine-tune their strategy for maximizing incremental sales. However, it also introduces entirely new themes to the social media and search conversation, and new attributes that are relevant to both the new product and others in its category.

  1. Beyond brand tracking: the broader potential 

Guiding and optimizing the launch of new products is just one example of how predictive big data can deliver benefits far outside the traditional brand-tracking arena. The same techniques that can predict brand equity can provide crucial real-time insight as to how a new product’s features meet the expectations of the market, the improvements that could drive loyalty and repeat purchase, and the likely impact of the launch on the brand owner’s current portfolio. And they can do so in a timeframe that enables effective launch optimization. When it comes to customer experience, effective social media listening is a vital tool for detecting divergences from the optimal, and feeding actionable insight on appropriate responses to the frontline. When we combine these new forms of real-time data with our growing understanding of situational equity, we are able to develop exception-based reporting that can provide early warning of the impact of negative headlines – or competitor activity.

  1. Final call for surveys? 

Does this mean the end for survey-based data? Far from it. We believe that a new search and social spine for brand tracking frees surveys from a future doing a job that they were never really suited to. And it unlocks the potential of shorter, smarter questionnaires to deliver new levels of understanding and value.

It’s a well-known fact that long surveys crammed with attribute-related questions are a disaster for data quality. Now that we are able to detect those attributes and predict brand equity faster and more cost-effectively through search and social, we can free our surveys from them. Instead we can target questionnaires at the areas where they can make the greatest contribution: providing individual-based insight that complements and deepens the aggregate view provided through search and social data.

  1. Introducing Intelligent, Adaptive Tracking 

Intelligent, Adaptive Tracking is the result: the integration of a real-time search and social view with in-depth surveys that can be triggered and served automatically whenever the listening data calls for it. When this data detects significant or unexpected variations in brand equity, surveys can be automatically triggered to help explore the reasons for them. When new themes are detected, we can use focused questionnaires to investigate further and reveal the role that these emerging discussions are playing within the category landscape.

The value of our search and social media spine increases exponentially the more we are able to attach responsive modules (the nerves and sinews of research) to it. By bringing our understanding of brand equity into the real-time present, we give ourselves and our clients the reaction time to explore in greater depth, develop more intelligent responses and make better decisions. The ability to do so can rejuvenate research’s contribution to brand management.